Morten Jerven is an economic historian from the LSE, with a good track record of publication and some World Bank and UNDP consultancies to his name. This book got the nod from Bill Gates and several good reviews. It makes a strong and succinct case for NOT relying on African GDP statistics as indicators of growth. The data from most sub-saharan countries are unreliable and misleading. One reason is that the aims of those who produce the figures and those who use them are in conflict. Jerven points out that scholars in the 1960s-1980s relied on national accounts, but by the 1990-2000s, Penn World Tables and World Bank data dominated footnotes; their ‘brand’ was better, although the ingredients were the same tainted data. Development studies, Jerven says, are now dominated by economists who prefer econometric analysis using global datasets in cross-country regressions; they are more interested in economics than economies, so they don’t notice the poor source data for the big data sets. The heart of Jerven’s argument rests (Ch 3) on case studies in which he dissects contradictory data on basic variables: population, agricultural production, and change in national income. The process of counting population in Nigeria is fraught with practical and ideological problems. Agricultural production figures have not adequately accounted for subsistence production--a major component (see sources on the informal economy, which dwarfs the formal economy). GDP and rates of change show huge variation from different sources. The reason for these basic data problems can be found in the bureaucracies of national capitals - civil servants without the tools to do what is expected, lack of investment in basic surveys and data collection, and poor institutions. “It requires a massive exercise of social power to establish valid numbers.” (Porter, 1995).
This book changed the way I think about aggregate data; it might NOT be getting better all the time. Canada’s experience with the long-form census, and the general retreat from big government science may mean that the world’s data are becoming less reliable rather than the reverse.
(also posted on Amazon)
David Last, 22 May 2013